1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
|
// Copyright 2014 The Gogs Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package charset
import (
"bytes"
"fmt"
"io"
"strings"
"unicode/utf8"
"code.gitea.io/gitea/modules/log"
"code.gitea.io/gitea/modules/setting"
"code.gitea.io/gitea/modules/util"
"github.com/gogs/chardet"
"golang.org/x/net/html/charset"
"golang.org/x/text/transform"
)
// UTF8BOM is the utf-8 byte-order marker
var UTF8BOM = []byte{'\xef', '\xbb', '\xbf'}
type ConvertOpts struct {
KeepBOM bool
}
// ToUTF8WithFallbackReader detects the encoding of content and converts to UTF-8 reader if possible
func ToUTF8WithFallbackReader(rd io.Reader, opts ConvertOpts) io.Reader {
buf := make([]byte, 2048)
n, err := util.ReadAtMost(rd, buf)
if err != nil {
return io.MultiReader(bytes.NewReader(MaybeRemoveBOM(buf[:n], opts)), rd)
}
charsetLabel, err := DetectEncoding(buf[:n])
if err != nil || charsetLabel == "UTF-8" {
return io.MultiReader(bytes.NewReader(MaybeRemoveBOM(buf[:n], opts)), rd)
}
encoding, _ := charset.Lookup(charsetLabel)
if encoding == nil {
return io.MultiReader(bytes.NewReader(buf[:n]), rd)
}
return transform.NewReader(
io.MultiReader(
bytes.NewReader(MaybeRemoveBOM(buf[:n], opts)),
rd,
),
encoding.NewDecoder(),
)
}
// ToUTF8 converts content to UTF8 encoding
func ToUTF8(content []byte, opts ConvertOpts) (string, error) {
charsetLabel, err := DetectEncoding(content)
if err != nil {
return "", err
} else if charsetLabel == "UTF-8" {
return string(MaybeRemoveBOM(content, opts)), nil
}
encoding, _ := charset.Lookup(charsetLabel)
if encoding == nil {
return string(content), fmt.Errorf("Unknown encoding: %s", charsetLabel)
}
// If there is an error, we concatenate the nicely decoded part and the
// original left over. This way we won't lose much data.
result, n, err := transform.Bytes(encoding.NewDecoder(), content)
if err != nil {
result = append(result, content[n:]...)
}
result = MaybeRemoveBOM(result, opts)
return string(result), err
}
// ToUTF8WithFallback detects the encoding of content and converts to UTF-8 if possible
func ToUTF8WithFallback(content []byte, opts ConvertOpts) []byte {
bs, _ := io.ReadAll(ToUTF8WithFallbackReader(bytes.NewReader(content), opts))
return bs
}
// ToUTF8DropErrors makes sure the return string is valid utf-8; attempts conversion if possible
func ToUTF8DropErrors(content []byte, opts ConvertOpts) []byte {
charsetLabel, err := DetectEncoding(content)
if err != nil || charsetLabel == "UTF-8" {
return MaybeRemoveBOM(content, opts)
}
encoding, _ := charset.Lookup(charsetLabel)
if encoding == nil {
return content
}
// We ignore any non-decodable parts from the file.
// Some parts might be lost
var decoded []byte
decoder := encoding.NewDecoder()
idx := 0
for {
result, n, err := transform.Bytes(decoder, content[idx:])
decoded = append(decoded, result...)
if err == nil {
break
}
decoded = append(decoded, ' ')
idx = idx + n + 1
if idx >= len(content) {
break
}
}
return MaybeRemoveBOM(decoded, opts)
}
// MaybeRemoveBOM removes a UTF-8 BOM from a []byte when opts.KeepBOM is false
func MaybeRemoveBOM(content []byte, opts ConvertOpts) []byte {
if opts.KeepBOM {
return content
}
if len(content) > 2 && bytes.Equal(content[0:3], UTF8BOM) {
return content[3:]
}
return content
}
// DetectEncoding detect the encoding of content
func DetectEncoding(content []byte) (string, error) {
// First we check if the content represents valid utf8 content excepting a truncated character at the end.
// Now we could decode all the runes in turn but this is not necessarily the cheapest thing to do
// instead we walk backwards from the end to trim off a the incomplete character
toValidate := content
end := len(toValidate) - 1
if end < 0 {
// no-op
} else if toValidate[end]>>5 == 0b110 {
// Incomplete 1 byte extension e.g. © <c2><a9> which has been truncated to <c2>
toValidate = toValidate[:end]
} else if end > 0 && toValidate[end]>>6 == 0b10 && toValidate[end-1]>>4 == 0b1110 {
// Incomplete 2 byte extension e.g. ⛔ <e2><9b><94> which has been truncated to <e2><9b>
toValidate = toValidate[:end-1]
} else if end > 1 && toValidate[end]>>6 == 0b10 && toValidate[end-1]>>6 == 0b10 && toValidate[end-2]>>3 == 0b11110 {
// Incomplete 3 byte extension e.g. 💩 <f0><9f><92><a9> which has been truncated to <f0><9f><92>
toValidate = toValidate[:end-2]
}
if utf8.Valid(toValidate) {
log.Debug("Detected encoding: utf-8 (fast)")
return "UTF-8", nil
}
textDetector := chardet.NewTextDetector()
var detectContent []byte
if len(content) < 1024 {
// Check if original content is valid
if _, err := textDetector.DetectBest(content); err != nil {
return "", err
}
times := 1024 / len(content)
detectContent = make([]byte, 0, times*len(content))
for i := 0; i < times; i++ {
detectContent = append(detectContent, content...)
}
} else {
detectContent = content
}
// Now we can't use DetectBest or just results[0] because the result isn't stable - so we need a tie break
results, err := textDetector.DetectAll(detectContent)
if err != nil {
if err == chardet.NotDetectedError && len(setting.Repository.AnsiCharset) > 0 {
log.Debug("Using default AnsiCharset: %s", setting.Repository.AnsiCharset)
return setting.Repository.AnsiCharset, nil
}
return "", err
}
topConfidence := results[0].Confidence
topResult := results[0]
priority, has := setting.Repository.DetectedCharsetScore[strings.ToLower(strings.TrimSpace(topResult.Charset))]
for _, result := range results {
// As results are sorted in confidence order - if we have a different confidence
// we know it's less than the current confidence and can break out of the loop early
if result.Confidence != topConfidence {
break
}
// Otherwise check if this results is earlier in the DetectedCharsetOrder than our current top guess
resultPriority, resultHas := setting.Repository.DetectedCharsetScore[strings.ToLower(strings.TrimSpace(result.Charset))]
if resultHas && (!has || resultPriority < priority) {
topResult = result
priority = resultPriority
has = true
}
}
// FIXME: to properly decouple this function the fallback ANSI charset should be passed as an argument
if topResult.Charset != "UTF-8" && len(setting.Repository.AnsiCharset) > 0 {
log.Debug("Using default AnsiCharset: %s", setting.Repository.AnsiCharset)
return setting.Repository.AnsiCharset, err
}
log.Debug("Detected encoding: %s", topResult.Charset)
return topResult.Charset, err
}
|
